Target Tracking-Oriented Detection Technology In Wireless Sensor Network

Wireless sensor network with the characteristics of low consumption,low cost,flexibility and miniaturization integrates data collection and communication,which is emerging discipline of the last century and changes the traditional data collection and communication mode.Therefore,it is evaluated as one of the most influential technologies in the 21 st century and becomes a hot research field in the current intelligent information system,widely applied in numerous industries including industrial,agriculture and smart home.One of the most significant applications of wireless sensor networks is to monitor and track target at the given monitored area,where the node position has an important influence on the data acquisition,especially for target tracking.Obviously,the accuracy of node position affects the performance of target location and tracking to a large extent,while the data collected by network without node location is meaningless.Besides,target attribute is also indispensable due to as a result of a fact that target attribute determines whether to start the target tracking mechanism.In this sense,target recognition occupies a huge place in tracking system,which not only affects tracking results,but also prolongs network life and improves system performance.Apparently,node localization and target recognition are two critical links in target tracking.Based on these two aspects,this paper proposes improved algorithms for the existing problems and the experimental data shows that the algorithm can significantly improve the network performance.The main research contents and innovations are described as bellow:(1)Against the current situation of node localization in wireless sensor network,this paper briefly outlines the technology of node self-localization and analyzes multi-dimensional scaling(MDS)localization algorithm detailedly in simulation aspects that consists of the influences of communication radius and distance measurement on localization based on two typical topological regions S and C type.Through the analysis of the simulation results,two main factors affecting node position in the multidimensional scale localization technology are obtained,including the distance estimation between nodes and the parameter optimization.So,an improved MDS algorithm on the basis of matrix correction and chaotic PSO is presented.The proposed algorithm firstly calculates the distance between the nodes by a recursive strategy,and adopts received signal strength to correct the distance,so as to reduce the distance error and avoid the network void problem.Then the chaotic particle swarm optimization algorithm is utilized to optimize the coordinate transformation parameters which can further decrease the influence of the parameters in coordinate transformation.Compared with similar algorithm,simulation experimental data shows that the algorithm can significantly improve the location performance.(2)Target feature extraction and selection is one of the key technologies in target recognition.To extract essential feature better,the feature extraction techniques are summarized,mainly presenting in summarizing different feature extraction technologies in time domain or frequency domain and pointing out advantages or disadvantages as well as the application scope.According to the comparison,wavelet is chosen as the basis to extract combined target feature containing optimal wavelet packet energy features and channel features.In order to decrease the complexity of the algorithm,reduce the computational amount and improve the accuracy of target recognition,the advantages of principal component analysis and linear discriminant analysis are combined to reduce the dimension of combinatorial features and discard redundant information.Through the data acquisition from real world,the compared results show that the technology has faster calculation speed and higher recognition rate.